Global Patterns of Human DNA Sequence Variation in a 10-kb Region on
Chromosome 1
Ning Yu,* Z. Zhao,† Y.-X. Fu,† N. Sambuughin,‡ M. Ramsay,§ T. Jenkins,§ E. Leskinen,\
L. Patthy,¶ L. B. Jorde,** T. Kuromori,* and W.-H. Li*
*Department of Ecology and Evolution, University of Chicago; †Human Genetics Center, University of Texas at Houston;
‡Neurology Research, Phoenix, Arizona; §Department of Human Genetics, South African Institute for Medical Research,
Johannesburg, South Africa; \Department of Biology, University of Oulu, Finland; ¶Institute of Enzymology, Hungarian
Academy of Sciences, Budapest, Hungary; and **Department of Human Genetics, University of Utah
Human DNA variation is currently a subject of intense research because of its importance for studying human
origins, evolution, and demographic history and for association studies of complex diseases. A ;10-kb region on
chromosome 1, which contains only four small exons (each ,155 bp), was sequenced for 61 humans (20 Africans,
20 Asians, and 21 Europeans) and for 1 chimpanzee, 1 gorilla, and 1 orangutan. We found 52 polymorphic sites
among the 122 human sequences and 382 variant sites among the human, chimpanzee, gorilla, and orangutan
sequences. For the introns sequenced (8,991 bp), the nucleotide diversity (p) was 0.058% among all sequences,
0.076% among the African sequences, 0.047% among the Asian sequences, and 0.045% among the European
sequences. A compilation of data revealed that autosomal regions have, on average, the highest p value (0.091%),
X-linked regions have a somewhat lower p value (0.079%), and Y-linked regions have a very low p value (0.008%).
The lower polymorphism in the present region may be due to a lower mutation rate and/or selection in the gene
containing these introns or in genes linked to this region. The present region and two other 10-kb noncoding regions
all show a strong excess of low-frequency variants, indicating a relatively recent population expansion. This region
has a low mutation rate, which was estimated to be 0.74 3 1029 per nucleotide per year. An average estimate of
;12,600 for the long-term effective population size was obtained using various methods; the estimate was not far
from the commonly used value of 10,000. Fu and Li’s tests rejected the assumption of an equilibrium neutral
Wright-Fisher population, largely owing to the high proportion of low-frequency variants. The age of the most
recent common ancestor of the sequences in our sample was estimated to be more than 1 Myr. Allowing for some
unrealistic assumptions in the model, this estimate would still suggest an age of more than 500,000 years, providing
further evidence for a genetic history of humans much more ancient than the emergence of modern humans. The
fact that many unique variants exist in Europe and Asia also suggests a fairly long genetic history outside of Africa
and argues against a complete replacement of all indigenous populations in Europe and Asia by a small Africa
stock. Moreover, the ancient genetic history of humans indicates no severe bottleneck during the evolution of humans
in the last half million years; otherwise, much of the ancient genetic history would have been lost during a severe
bottleneck. We suggest that both the ‘‘Out of Africa’’ and the multiregional models are too simple to explain the
evolution of modern humans.
Introduction
Human DNA variation is currently a subject of intense research for several reasons. First, DNA variation
within and between human populations is of great interest to human geneticists and evolutionists. Second,
there is much interest in the utility of single nucleotide
polymorphisms (SNPs) in molecular medicine, because
SNP markers may be useful in association studies of
complex diseases, assessment of individuals’ predisposition to diseases, and tailoring of therapies. Third, the
availability of long genomic sequences generated by the
Human Genome Project and the advent of inexpensive
DNA sequencing techniques have made large-scale population studies feasible. In fact, there are now many
large-scale population studies of human DNA variation.
These include regions containing the genes for, respectively, b-globin (Harding et al. 1997), lipoprotein lipase
(Nickerson et al. 1998), angiotensin-converting enzyme
(Rieder et al. 1999), and pyruvate dehydrogenase E1
Key words: nucleotide diversity, DNA variation, human evolution, unique variants.
Address for correspondence and reprints: Wen-Hsiung Li, Department of Ecology and Evolution, University of Chicago, 1101 East
57th Street, Chicago, Illinois 60637. E-mail: [email protected].
Mol. Biol. Evol. 18(2):214–222. 2001
q 2001 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038
214
(PDHA1) (Harris and Hey 1999) and an 8-kb intron segment of the dystrophin gene (Zietkiewicz et al. 1998),
a 10-kb noncoding region in Xq13.3 (Kaessmann et al.
1999), the replication origin IR and the flanking and
coding regions of the b-globin gene (Fullerton et al.
2000), and a 10-kb noncoding region in 22q11.2 (Zhao
et al. 2000). To date, however, the number of large-scale
worldwide surveys of DNA sequence variation is still
small, especially for noncoding regions.
We have been pursuing human DNA variation studies in noncoding regions for two purposes. First, we
wish to establish a genomewide and worldwide neutrality standard of nucleotide diversity. By ‘‘neutrality standard,’’ we mean the level of nucleotide diversity expected in a region in which all mutations are neutral and
not directly subject to natural selection. This standard
will be a very useful reference, especially for comparison with the levels of nucleotide diversity in coding regions. Obviously, such a standard requires data from
many genomic regions, because the level of nucleotide
diversity in a region is subject to strong stochastic effects. Second, we wish to study the origin and evolution
of modern humans. DNA sequence data from noncoding
regions may more accurately reflect human history than
data from coding regions, because noncoding regions
Global Pattern of Human DNA Variation
215
are not directly subject to natural selection. The majority
of past studies on human DNA variation, which are
mainly from mitochondrial (mt) DNA, microsatellite
DNA, and the Y chromosome, have largely given the
impression of a relatively shallow genetic history of humans. These observations have been taken as evidence
for the Out of Africa model for the origin of modern
humans, which postulates that a founder group of modern humans emigrated from Africa about 100,000 years
ago to Europe and Asia and completely replaced all the
indigenous populations outside of Africa (Cann, Stoneking, and Wilson 1987; Stringer and Andrew 1988).
However, recent studies of the b-globin and the PDHA1
gene regions (Harding et al. 1997; Harris and Hey 1999)
and a 10-kb noncoding region on chromosome 22 (Zhao
et al. 2000) have revealed an ancient genetic history of
humans and suggested that human evolution has been
more complex than depicted by the simple Out of Africa
model. To attain a better understanding of this issue, it
is necessary to obtain sequence data from other noncoding regions.
For the above purposes, we selected a ;10-kb region on chromosome 1 and obtained sequence data from
worldwide populations. This region contains mostly introns, although it also includes four short exons. The
new data were compared with the data from Xq13.3 and
22q11.2 and other regions to study the features of sequence variation within and between populations and
were used to infer the genetic history of human
evolution.
two overlapping fragments covering positions 6880–
11022 in the 12-kb region. Touchdown PCR (Don et al.
1991) was used, and the reactions were carried out under
the conditions described in Zhao et al. (2000). The PCR
products were purified with the Wizard PCR Preps DNA
Purification Resin Kit (Promega). Sequencing reaction
was performed according to the protocol of ABI Prism
BigDye Terminator Sequencing Kits (Perkin Elmer)
modified by quarter reaction. The extension products
were purified by Sephadex G-50 (DNA grade, Pharmacia) and run on an ABI 377XL DNA sequencer using
4.25% gels (Sooner Scientific).
ABI DNA Sequence Analysis 3.0 was used for lane
tracking and base calling. The data were then proofread;
the fluorescence traces were reread manually and heterozygous sites were detected as double peaks. The segment sequences were assembled automatically using
SeqMan in DNASTAR. The assembled files were carefully checked manually using the same program, and
variant sites were identified in the aligned sequences in
MegAlign in DNASTAR. All of the nucleotides in each
segment were sequenced at least once in both directions.
Furthermore, all singletons and doubletons, which are
defined as variants that appear, respectively, only once
and twice in the total sample, were verified by reamplifying the region containing the variant site and resequencing the region in both directions using new internal
primers that were close to the site.
Materials and Methods
Region Selection and Populations Sampled
The sequences were aligned by MegAlign in the
DNASTAR software package. The human consensus sequence was obtained from the alignment using DNASTAR. The human ancestral sequence was inferred by
comparing the human sequences with the outgroup sequences using the maximum-parsimony principle.
For a DNA sequence subject to no natural selection, the mutation rate per sequence per generation (m)
is estimated by
A 12-kb region corresponding to nucleotide positions 18333–30332 in locus HS125H23 on human chromosome 1q24 was selected (GenBank accession number
Z94054) because no gene was registered in this region
at GenBank. After excluding a region containing a
polyA segment and a region containing an MER33, an
incomplete AluSx repeat, and an MIR2 repeat, a total of
;10 kb of nucleotide sites were selected for sequencing.
Initially, no potential coding region was detected in the
12-kb region by XGRAIL. However, upon reexamination, three strong potential coding regions (exons) and
a weak potential coding region were detected by both
GenScan and GRAIL-EXP (see Results). As these potential exons are short (#155 bp), the region selected
covers largely introns.
Sixty-one individuals were collected worldwide
from 14 human populations in three major geographic
areas: 20 Africans (5 South African Bantu speakers, 1
!Kung, 2 Mbuti Pygmies, 2 Biaka Pygmies, 5 Nigerians,
5 Kenyans), 20 Asians (8 Chinese, 3 Japanese, 6 Indians, 3 Yakuts), and 21 Europeans (6 Swedes, 2 Finns,
5 French, 5 Hungarians, 3 Italians). One chimpanzee,
one gorilla, and one orangutan were used as outgroups.
PCR Amplification and DNA Sequencing
Five primer pairs were designed to amplify three
overlapping fragments covering positions 9–6141 and
Data Analysis
m 5 dgL/(2t),
(1)
where d is the number of nucleotide substitutions per
site between two sequences, t is the divergence time
between two sequences, L is the sequence length (bp),
and g is the generation time (g ø 20 years for humans).
Watterson’s (1975), Tajima’s (1983), and Fu’s (1994)
methods were used to estimate u 5 4Nem, where Ne is
the effective population size.
Tajima’s (1989) test and Fu and Li’s (1993) tests were
used to test the selective neutrality of the region studied;
a program is available at http://hgc.sph.uth.tmc.edu/fu. The
critical points (values) for the neutrality tests were obtained
from 5,000 simulated samples. Fu’s (1996) and Fu and Li’s
(1997) methods were used to estimate the age of the most
recent common ancestor (MRCA) of the DNA sequences
in a sample. We computed both the mode and mean of
the age (T) of the MRCA in years.
Note that all of the above computations require
only segregating site data but do not require haplotype
data.
216
Yu et al.
Table 1
Numbers of Variant Sizes (excluding indels) in each DNA Region Among Sequences
10
KB IN
1Q24
10
KB IN
22Q11.2a
10
KB IN
XQ13.3b
TYPE OF
VARIANT
All
(122)c
Africans
(40)
NonAfricans
(82)
All
(128)
Africans
(40)
NonAfricans
(88)
All
(69)
Africans
(23)
NonAfricans
(46)
Singleton . . . .
Doubleton. . . .
Others . . . . . . .
Total . . . . . . . .
19 (3)d
7 (0)
22 (1)
48 (4)
7 (3)
2 (0)
20 (1)
29 (4)
12 (0)
5 (0)
11 (0)
28 (0)
20 (2)
23 (1)
32 (1)
75 (4)
8 (1)
19 (0)
27 (1)
54 (2)
12 (1)
4 (1)
28 (0)
44 (2)
19 (0)
5 (0)
9 (0)
33 (0)
11 (0)
5 (0)
8 (0)
24 (0)
8 (0)
0 (0)
9 (0)
17 (0)
a
From Zhao et al. (2000).
From Kaessmann et al. (1999).
Number of sequences studied.
d The numbers of indels are given in parentheses.
b
c
Results and Discussion
Sequence Data
We sequenced ;9,626 nucleotide sites in the selected region in 61 humans (AF310265–AF310325), 1
chimpanzee (AF310683), 1 gorilla (AF310682), and 1
orangutan (AF310681). The human consensus and ancestral sequences were obtained as explained in Materials and Methods. The GC contents of the human consensus, human ancestral, chimpanzee, gorilla, and
orangutan sequences were ;31.5%, which is much lower than the genome average of 42%. Thus, the region
studied was GC-poor.
Three strong potential coding regions (exons) in the
selected 12-kb segment were predicted by GenScan
(each with a probability .95.5%) and GRAIL-EXP;
these potential exons were at sites 23522–23632,
26131–26285, and 27542–27679 in locus HS125H23
(Z94054), respectively. In fact, a BLAST search showed
that the amino acid sequence translated from these three
exons was 86% similar to a segment of a human membrane protein CH1 (GenBank accession number
AF097535). Later on a BLAST search of the GenBank
with the nucleotide sequence of the ;10 kb region indicated that four exons are similar to the human membrane protein CH1 (similarity . 99%) and the amino
acid sequence translated from these four exons is identical to that translated from this gene. These exons are
from site 21,773 to 21,888, site 26,128 to 26,285, site
27,542 to 27,679, and site 27,902 to 28,056 in locus
HS125H23 (Z94054), respectively.
Pattern of Sequence Variation
A total of 48 variant sites were found in the alignment of human sequences; 19 of them were observed
only once (i.e., singletons), 7 were observed twice (i.e.,
doubletons), and 22 were observed more than twice (i.e.,
others) (table 1). Two variant sites (one synonymous and
one nonsynonymous singleton) were observed in the
second exon, while all of the remaining 46 variant sites
were found in introns. All singletons and doubletons
were verified as explained in Materials and Methods,
and no error was found. In addition to the 48 singlenucleotide variants, we found 4 insertions/deletions (indels) among the 122 human sequences. On average, ;5
variant sites per 1,000 bp were found in the region
studied.
The numbers of variant sites (excluding indels) in
the African, Asian, and European sequences were 29,
20, and 16, respectively (table 1). Thus, Africans had
the largest number of variants. The pattern of sequence
variation in Africans differed somewhat from that in
non-Africans. For example, less than one third of the
variant sites in Africans were singletons, whereas close
to one half of the variant sites in non-Africans were
singletons. Interestingly, the proportions of unique variant sites in each of the three continents were high: 20
(69%, including 7 singletons), 11 (55%, 8 singletons),
and 8 (50%, 4 singletons) among the African, Asian,
and European sequences, respectively. This observation
suggests a substantial degree of isolation between
continents.
Table 1 also includes the patterns of sequence variation in the 10-kb noncoding regions on chromosomes
X and 22 (Kaessmann et al. 1999; Zhao et al. 2000).
Note that among the African sequences, the number of
low-frequency variants (i.e., singletons and doubletons)
was smaller than that of high-frequency variants (i.e.,
others) in the present region, whereas the opposite was
true for the two other regions. On the other hand, among
non-African sequences, the number of low-frequency
variants was larger than that of the high-frequency variants in the present region, whereas the opposite was
true for the other two regions. Thus, although a stronger
excess of low-frequency variants in Africans than in
non-Africans was observed in the previous two regions,
the opposite was found in the present region. This difference in pattern notwithstanding, two features common to the three regions were noted. First, there were
more variants in the African sample than in the nonAfrican sample, despite the number of Africans studied
being less than half that of non-Africans studied. Thus,
Africans were considerably more polymorphic than nonAfricans, in agreement with previous observations
(Cann, Stoneking, and Wilson 1987; Kaessmann et al.
1999; Zhao et al. 2000). Second, the number of lowfrequency variants (singletons and doubletons) in the total sample was larger than the number of high-frequency
variants, e.g., 26 versus 22 in the present region. This
excess of low-frequency variants in all three regions is
Global Pattern of Human DNA Variation
217
Table 2
Distribution of the Numbers of Variant Sites Observed in the Human and Outgroup Sequences
TYPE OF
VARIATION
SUBREGIONS (;1 kb each)
1
2
3
4
5
6
7
8
9
10
Total
36
5
41
4
30
4
34
6
35
7
42
6
32
8
40
9
38
4
42
6
35
3
38
3
28
1
29
2
21
1
22
2
338
44
382
52
10-kb region in 22q11.2 (Zhao et al. 2000)
Substitutions . .
37
36
33
Indels . . . . . . . .
2
14
10
Total. . . . . . . . .
39
50
43
51
2
53
41
4
45
45
5
50
40
3
43
48
2
50
49
2
51
36
3
39
416
47
463
10-kb region in Xq13.3 (Kaessmann et al. 1999)
Substitutions . .
20
29
20
Indels . . . . . . . .
1
5
2
Total. . . . . . . . .
21
34
22
16
0
16
18
1
19
20
2
22
26
0
26
14
3
17
29
2
31
24
0
24
216
16
232
10-kb region in 1q24 (present study)
Substitutions . .
41
42
Indels . . . . . . . .
5
6
Total. . . . . . . . .
46
48
Human . . . . . . .
8
6
in sharp contrast to the situations for the dystrophin and
PDHA1 genes (Zietkiewicz et al. 1998; Harris and Hey
1999) and suggests a relatively recent population expansion, because such an excess is not expected from
an equilibrium Wright-Fisher population.
The present region was considerably less polymorphic than the region on chromosome 22—it had fewer
high-frequency variants, especially among non-Africans,
and a much smaller number of doubletons. The relatively low polymorphism may be due to a lower mutation rate (see below) and selection in the gene containing
these introns or in genes linked to this region. The Xlinked region was even less variable (table 1). This may
be because an X-linked region has a smaller effective
population size than an autosomal region (3Ne/4 vs. Ne)
and because the X region has a low recombination rate
(Kaessmann et al. 1999), so that compared to the other
two regions, it is subject to stronger background selection (i.e., effects of deleterious mutations in genes linked
to the region) or selective sweep (effects of positive selection in genes linked to the region) (Begun and
Aquadro 1992; Charlesworth 1994).
A comparison of all sequences, including the chimpanzee, gorilla, and orangutan sequences, revealed 382
Table 3
Mutation Pattern
Direction of Mutation
Can Be Inferred
A→C
A→G
A→T
C→A
C→G
C→T
G→A
G→C
G→T
T→A
T→C
T→G
8
42
7
9
11
21
23
5
4
2
27
13
Subtotal . . . . . . . . . . .172
Transitions (%) . . . . .113 (66%)
Direction of Mutation Cannot
Be Inferred
A↔G
C↔T
A↔C
A↔T
G↔C
G↔T
66
44
14
17
12
16
169
110 (65%)
variant sites, 44 of which were indels. The 382 variant
sites were evenly distributed in this region (x2 5 14.4,
df 5 9, P 5 0.11) (table 2). The number of variant sites
in human populations was also evenly distributed (x2 5
9.9, df 5 9, P 5 0.36).
Among the 44 indels in our data, 13, 7, 10, and 10
were 1-, 2-, 3-, and 4-nt indels, respectively. The remaining 4 indels involved 5, 8, or .10 nt. Therefore,
the majority of these indels were short.
Mutation Pattern
Comparing the human, chimpanzee, gorilla, and
orangutan sequences, we were able to infer the direction
of 172 mutations (table 3); the proportion of transitional
changes was 66%. For the 169 mutations for which the
direction could not be inferred, the proportion of transitions was 65% (table 3). This proportion was between
the values (59% and 70%) observed in pseudogenes (Li,
Wu, and Luo 1984) and for the 10-kb region in 22q11.2
(Zhao et al. 2000). For those mutations whose direction
could be inferred, the number of G/C-to-A/T mutations
was 57, while that of A/T-to-G/C mutations was 90. According to the GC and AT contents of 68.5% and 31.5%,
the expected numbers of G/C-to-A/T mutations and A/
T-to-G/C mutations are 100.7 and 46.3, respectively, and
a comparison with the observed numbers gives x2 5
3.61 and P 5 0.057, which is close to significant. This
result suggests that G/C-to-A/T mutations might occur
more frequently than A/T-to-G/C mutations, similar to
the situation for mammalian pseudogenes (G/C-to-A/T,
64.5%; A/T-to-G/C, 35.5%) (Li 1997).
Neutrality Tests
The assumption that the region under study is subject to no natural selection was tested using the sequence
data. Using the critical points obtained from the 5,000
samples we simulated, we found that Tajima’s and Fu
and Li’s tests could not reject the neutral Wright-Fisher
model when each continent was considered separately
(table 4). However, when we used data from more than
one continent, the results became different. Although
218
Yu et al.
Table 4
Neutrality Tests
No. of
Sequences
Test
u
Test Value
Critical Value
(P 5 0.05)
P
Asian sequences
Tajima’s test . . . . . . . . . . . . . . . . . . .
Fu and Li’s D . . . . . . . . . . . . . . . . . .
Fu and Li’s F . . . . . . . . . . . . . . . . . .
40
40
40
4.26
4.26
4.26
20.293
21.230
20.997
21.43
21.86
21.78
.0.10
.0.10
.0.10
European sequences
Tajima’s test . . . . . . . . . . . . . . . . . . .
Fu and Li’s D . . . . . . . . . . . . . . . . . .
Fu and Li’s F . . . . . . . . . . . . . . . . . .
42
42
42
4.14
4.14
4.14
0.145
20.021
0.042
21.39
21.91
21.78
.0.10
.0.10
.0.10
African sequences
Tajima’s test . . . . . . . . . . . . . . . . . . .
Fu and Li’s D . . . . . . . . . . . . . . . . . .
Fu and Li’s F . . . . . . . . . . . . . . . . . .
40
40
40
6.82
6.82
6.82
0.000
20.051
20.036
21.43
21.84
21.76
.0.10
.0.10
.0.10
Non-African sequences
Tajima’s test . . . . . . . . . . . . . . . . . . .
Fu and Li’s D . . . . . . . . . . . . . . . . . .
Fu and Li’s F . . . . . . . . . . . . . . . . . .
82
82
82
4.17
4.17
4.17
20.768
22.226*
21.889*
21.43
21.90
21.77
.0.10
0.02 , P , 0.03
0.04
All sequences
Tajima’s test . . . . . . . . . . . . . . . . . . .
Fu and Li’s D . . . . . . . . . . . . . . . . . .
Fu and Li’s F . . . . . . . . . . . . . . . . . .
122
122
122
5.29
5.29
5.29
21.224
22.601*
22.325*
21.41
21.86
21.70
0.05 , P , 0.10
0.01 , P , 0.02
0.01 , P , 0.02
All sequences (including indels)
Tajima’s test . . . . . . . . . . . . . . . . . . .
Fu and Li’s D . . . . . . . . . . . . . . . . . .
Fu and Li’s F . . . . . . . . . . . . . . . . . .
122
122
122
5.83
5.83
5.83
21.200
22.994*
22.564*
21.38
21.79
21.64
0.05 , P , 0.10
0.005 , P , 0.01
0.005 , P , 0.01
NOTE.—The critical points (values) were obtained from 5,000 simulated samples. The indels were excluded in each test except for the last three tests.
* Significant at the 5% level.
Tajima’s test remained nonsignificant, Fu and Li’s tests
for non-Africans and for all samples were significant
(table 4); the conclusion became even stronger when
indels were included in the tests. The rejection of neutrality was largely due to the high proportion of lowfrequency variants. As the region studied was largely
noncoding, the rejection of neutrality may be due to two
factors: (1) a relatively recent population expansion,
which can increase the number of low-frequency variants, and (2) natural selection in the exons or in genes
linked to this region. In fact, the region is ;44 kb and
;110 kb away from a functional gene at its 59 and 39
ends, respectively.
As the rejection of the neutrality assumption was
largely due to the excess of low-frequency variants, one
might wonder whether the excess was due to pooling of
data from different populations. However, data pooling
actually tends to increase, rather than decrease, the proportion of low-frequency variants, as implied by the reTable 5
Expected Proportion of Singleton Mutations in the Total
Sample when 10 Sequences Are Sampled from each
Subpopulation
NO.
OF
SUBPOPULATIONS
2. . . . . . . . . . . . . .
4. . . . . . . . . . . . . .
10. . . . . . . . . . . . . .
0.1
1.0
RANDOM
MATING (no
subdivision)
0.074
0.017
0.004
0.210
0.100
0.036
0.281
0.235
0.190
MIGRATION RATE (4Nm)
sult of Fu (1996), who showed that Fu and Li’s D test
tends to be positive under population subdivision. Since
the D test compares the number of singletons with the
total number of mutations, Fu’s (1996) result indicates
that the proportion of singletons is reduced under population subdivision. To show this, we conducted a coalescent simulation using Wright’s island model. The expected number of singleton mutations was assumed to
be equal to the expected sum of lengths of external
branches in the sequence genealogy, which is expressed
in units of u 5 4Nem, where Ne is the effective size of
the entire population and m is the rate of mutation per
sequence per generation. The results are shown in table
5, in which each entry is the ratio of the expected sum
of length of external branches and the expected total tree
length of the simulated sequence genealogy; the ratio is
independent of u. Under the assumption of a single random mating population with the same u, the expected
sums of lengths of external branches and all branches
are 1 and 1 1 ½ 1 . . . 1 1/(n 2 1), respectively, where
n is the sample size. As can be seen from table 5, the
more subpopulations or the less migration, the lower the
proportion of singletons relative to the total number of
mutations in the sample. Clearly, population subdivision
tends to reduce the proportion of low-frequency
variants.
Nucleotide Diversity
Nucleotide diversity (p) is defined as the average
number of nucleotide differences per site between two
Global Pattern of Human DNA Variation
219
Table 6
Nucleotide Diversity Within and Between Populations in Noncoding Regions
NO. OF
SE-
TYPE OF
SEQUENCE
QUENCES
Introns
Noncoding
59 flanking
Introns
39 flanking
122
128
349
349
349
59 flanking
IR
39 flanking
36
36
36
X-linkedb
Xq13.3 . . . . . .
PDHA1 . . . . .
Dys44 . . . . . .
ZFX . . . . . . . .
Subtotal . . . . .
Noncoding
Introns
Introns
Introns
Y-linkedb
ZFY . . . . . . . .
SRY . . . . . . . .
Subtotal . . . . .
Introns
Introns
Autosomal. . . . .
X-linked . . . . . .
Alus
Alus
REGION
Autosomalb
1q24 . . . . . . . .
22q11. . . . . . .
b-globin . . . . .
Subtotal . . . . .
b-globin IR. .
NUCLEOTIDE DIVERSITY (%)
LENGTH
(bp)
Total
Africa
Asia
Europe
Africa/
Asia
Africa/
Europe
Asia/
Europe
0.045
0.077
0.232
0.116
0.255
0.076
0.029
0.159
0.147
0.065
0.083
0.393
0.171
0.304
0.097
0.063
0.108
0.309
0.096
0.242
0.099
0.104
0.290
0.140
0.046
0.091
0.331
0.182
0.319
0.092
0.034
0.009
0.071
0.000
0.041
0.038
0.248
0.107c
0.030c
0.096
0.034
0.251
0.110c
0.028c
0.096
0.035
0.005
0.084
0.002
0.045
REFER-
8,991
9,834
814
980
582
21,201
2,902
1,300
1,874
0.058
0.088
0.318
0.140
0.264
0.091
0.075
0.232
0.141
0.076
0.085
0.323
0.072
0.217
0.093
0.136
0.287
0.119
0.047
0.075
0.331
0.142
0.273
0.081
69
35
847
336
9,564
3,530
7,185
782
21,061
0.034
0.169
0.102
0.021
0.079
0.035
0.216
0.109c
0.053c
0.091
0.025
0
0.085
0.004c
0.040
205
5
676
;17,685
18,361
0.001
0.008
0.008
8
9
1,138
1,455
0.263
0.125
1, 10
4–7
ENCESa
1
2
2
2
3
3
3
4
5
6
7
a 1 5 Zhao et al. (2000); 2 5 Harding et al. (1997); 3 5 Fullerton et al. (2000); 4 5 Kaessmann et al. (1999); 5 5 Harris and Hey (1999); 6 5 Zietkiewicz
et al. (1998); 7 5 Jaruzelska et al. (1999); 8 5 Jaruzelska, Zietkiewicz, and Labuda (1999); 9 5 Whitfield, Sulston, and Goodfellow (1995); 10 5 Nickerson et
al. (1998).
b Alu sequences were excluded.
c African Americans were excluded from the analysis.
randomly chosen sequences from the population. The p
value was calculated using the program DNASP, version
3.0 (Rozas and Rozas 1999). For the introns sequenced,
the p value was 0.058% among all sequences, 0.076%
among the African sequences, 0.047% among the Asian
sequences, and 0.045% among the European sequences
(table 6). Thus, the p value was largest among Africans.
These p values were considerably lower than those for
the 10-kb noncoding region in 22q11.2 (Zhao et al.
2000) and the average value (0.11%) for fourfold-degenerate sites in coding genes in white Americans (Li
and Sadler 1991). The low nucleotide diversity in this
region may be due to a lower mutation rate (see below)
and/or background selection and selective sweep.
Table 6 presents a list of the p values for various
noncoding regions. The p values are usually highest in
Africans but are quite similar in Asians and Europeans.
For example, the average p values for the autosomal
regions are 0.093%, 0.081%, and 0.076% for the Africans, Asians, and Europeans, respectively. These values
are somewhat lower than the average p value (0.11%)
at fourfold-degenerate sites in 49 genes. However, the
number of noncoding regions studied is small, so no
general conclusion can be drawn yet.
A large variation in p is seen among regions (table
6). In particular, the 59 and 39 flanking regions of the bglobin gene and the b-globin replication origin initiation
region (IR) have high p values (Harding et al. 1977;
Fullerton et al. 2000). The high p value in the IR has
been speculated to be due to a high mutation rate because of the peculiar feature of the DNA unwinding element in the IR (Fullerton et al. 2000). On average, the
autosomal regions have the highest p value (0.091%),
the X-linked regions have a somewhat lower p value
(0.079%), and the Y-linked regions have a very low p
value (0.008%). These differences may be partly due to
the fact that the relative effective population sizes are
Ne, 3Ne/4, and Ne/4 for an autosomal, an X-linked, and
a Y-linked sequence, respectively. However, the extremely low p value for Y-linked sequences may be
mainly due to background selection and selective sweep,
because there is no recombination in the Y chromosome
except for the pseudoautosomal region. Background selection and selective sweep should have, on average,
stronger effects on an X-linked region than on an autosomal region because of a lower average recombination rate in the X chromosome than in an autosome,
partly accounting for the lower p value for X-linked
regions. The number of Alu sequences studied is small,
but the data suggest a higher average p value for Alus
than for other regions. This is not surprising, because
Alus have higher mutation rates due to the presence of
a high frequency of CpG dinucleotides.
Mutation Rate, u, Ne
The average numbers of nucleotide substitutions
per site were 0.62% between human and chimpanzee
220
Yu et al.
Table 7
Estimates of the Mutation Rate per Site per Year (v), Parameter u, and Effective Population Size (Ne)
ESTIMATED VALUES
PARAMETER
Chimpanzee vs. Human
Divergence
time (Myr) . . . . . .
(3109) .
v
.........
m (3104)a . . . . . . . . .
u (Ne 5 10,000)b . . .
Ne (W)c . . . . . . . . . . .
Ne (T)d . . . . . . . . . . .
Ne (BLUE)e . . . . . . .
Gorilla vs. Human
Orangutan vs. Human
5
6
7
7
8
9
12
14
16
0.62
1.11
4.46
19,200
11,800
14,100
0.52
0.93
3.72
23,000
14,100
16,900
0.44
0.80
3.19
26.800
16,500
19,700
0.76
1.37
5.50
15,500
9,600
11,500
0.67
1.20
4.81
17,800
10,900
13,100
0.59
1.07
4.28
20,000
12,300
14,700
1.02
1.83
7.31
11,700
7,200
8,600
0.87
1.57
6.27
13,600
8,400
10,000
0.76
1.37
5.48
15,600
9,600
11,500
m 5 mutation rate per sequence per generation, assuming 20 years per generation.
u was estimated as u 5 4Nem, with Ne 5 10,000.
c The N value estimated as u/4m using the u value (8.55) estimated by Watterson’s (1975) method.
e
d The N value estimated by u/4m using the u value (5.26) estimated by Tajima’s (1983) method.
e
e The N value estimated by u/4N using u value 6.30 estimated by Fu’s (1994) BLUE method.
e
a
b
sequences, 1.07% between human and gorilla sequences, and 2.44% between human and orangutan sequences;
here, we exclude the four exons. The mutation rates (v)
were estimated to be 0.52 3 1029, 0.67 3 1029, and
1.02 3 1029 per nucleotide site per year based on divergence times of 6 Myr between humans and chimpanzees, 8 Myr between humans and gorillas, and 12
Myr between humans and orangutans, respectively; other divergence dates are also considered in table 7. The
first two values are considerably smaller than the third,
but the differences may be largely due to stochastic fluctuations. The average for the three estimates is 0.74 3
1029. This value is considerably lower than the estimate
(1.15 3 1029) obtained from the 10-kb region on chromosome 22, but it is consistent with the lower nucleotide diversity in this region than in the 10-kb region on
chromosome 22. It is possible that this region has a lower (neutral) mutation rate.
Several methods are available for estimating the
parameter u 5 4Nem, where m 5 vgL, g is the generation
time (20 years), and L 5 8,991 bp (see eq. 1). If we use
the commonly used value 10,000 for Ne (Takahata
1993), u varies with the assumption of divergence dates
(table 7). For the average mutation rate obtained above,
we obtained u 5 5.32. For the two commonly used
methods, one known as Watterson’s (1975) estimator
and the other as Tajima’s (1983) estimator, we obtained
Table 8
The Age (T, 103 years) of the MRCA of the Human
Sequences Sampled
Sequences
Ne
Tmode
Tmean
95% Interval
All samples . . . .
10,000
12,000
15,000
6,000
8,000
10,000
6,000
7,000
8,000
1,472
1,238
1,080
1,061
922
832
677
655
672
1,503
1,383
1,229
1,096
1,039
987
805
779
757
728;2,392
643;2,314
564;2,136
523;1,795
480;1,766
448;1,712
341;1,445
330;1,422
320;1,382
Africans . . . . . . .
Non-Africans . . .
NOTE.—An average mutation rate of 1.33 3 1024 per sequence per generation was used.
u 5 8.55 and 5.26, respectively. Note that these two
methods are not optimal in terms of minimizing variance. The minimum variance estimator (BLUE) by Fu
(1994) gave u 5 11.50. BLUE and Watterson’s (1975)
estimator yielded larger u values, mainly because there
was an excess of singletons (table 1). To avoid the effect
of the excess of low-frequency variants, we excluded
singletons and doubletons from analysis and obtained
the BLUE estimate of u 5 6.30, which is similar to the
other two estimates.
If we know the mutation rate, we can estimate the
effective population size Ne from u. As the estimate of
the mutation rate varies with the assumption of the divergence dates, the estimate of Ne also varies (table 7).
Moreover, it also depends on the estimation methods
used (table 7). If we use the average mutation rate obtained above and the average (6.70) of the u values estimated by Watterson’s, Tajima’s, and the BLUE methods, we obtain Ne 5 12,600, which is not far from the
commonly used value of 10,000 (Takahata 1993).
Age of the MRCA
To estimate the age (T) of the MRCA of the sequences in a sample, the values of both Ne and mutation
rate per sequence per generation (u) are required. As
mentioned above, the estimate of mutation rate depends
on the species pair and the divergence dates used. For
simplicity, we shall use the average mutation rate obtained above, i.e., v 5 0.74 3 1029 changes per site per
year and u 5 1.33 3 1024 changes per sequence per
generation in humans; the estimate of T increases with
decreasing u. Table 8 presents the estimates of T for
several values of effective population sizes for the entire
sample, the subsample of sequences from Africa only,
and the subsample of non-African sequences only, respectively. If the commonly used Ne 5 10,000 was assumed, the mode estimate (Tmode) and mean estimate
(Tmean) were, respectively, 1,376,000 and 1,559,000
years for the entire sample. These estimates were comparable to our previous estimates based on the polymorphism data from a 10-kb region on chromosome 22
(Zhao et al. 2000). The estimates based on the African
Global Pattern of Human DNA Variation
sample were only somewhat smaller than those based
on the entire sample, while those based on the non-African sample were the smallest. This pattern was also
consistent with our previous study (Zhao et al. 2000).
It should be pointed out that the method used to
estimate the age of the MRCA makes the assumption
that the sample was taken from a random-mating population with a constant effective population size. However, the fact that there are many unique variants in each
of the three continents (see above) suggests that this
assumption may not be appropriate. The existence of
more singletons than expected would tend to inflate our
estimates of the age of the MRCA. Methods that make
a better use of segregating sites of different types are
needed for correcting such biases, and one of us is in
the process of developing such methods. Note that the
alternative method developed by Griffiths and Tavaré
(1994) is not applicable here because we do not have
haplotype sequences. Nevertheless, it is unlikely that the
revised estimates will be smaller than half of the values
given in table 8.
In summary, like the data from the PDHA1 locus
(Harris and Hey 1999) and the 10-kb region on chromosome 22 (Zhao et al. 2000), the present data also
provide evidence for a genetic history of humans that is
much more ancient than the emergence of modern humans. The observation that both the region on chromosome 22 and the present region show an ancient genetic history outside of Africa argues against a complete
replacement of all indigenous populations in Europe and
Asia by an African stock. Moreover, the ancient genetic
history of humans indicates no severe bottleneck during
the evolution of humans in the last half million years,
because much of the ancient genetic history would have
been lost during a severe bottleneck. On the other hand,
the fact that most available nuclear DNA variation data,
as well as mitochondrial DNA data, show a considerably
shallower genetic history in Asia and Europe than in
Africa suggests that human evolution has not occurred
in parallel in different parts of the Old World, as depicted by the multiregional model. Thus, both the Out
of Africa and the multiregional models appear to be too
simple to explain the evolution of modern humans.
Acknowledgments
We thank Drs. J. B. Clegg, Marie Lin, and Maryellen Ruvolo for DNA samples. This work was supported by NIH grants GM55759 (W.-H.L.), GM50428
(Y.-X.F.), GM59290 (L.B.J.) and NSF DEB9707567 (Y.X.F.).
LITERATURE CITED
BEGUN, D. J., and C. F. AQUADRO. 1992. Levels of naturally
occurring DNA polymorphism correlate with recombination
rates in D. melangaster. Nature 356:519–520.
CANN, R. L., M. STONEKING, and A. C. WILSON. 1987. Mitochondrial DNA and human evolution. Nature 325:31–36.
CHARLESWORTH, B. 1994. The effect of background selection
against deleterious alleles on weakly selected, linked variants. Genet. Res. 63:213–228.
221
DON, R. H., P. T. COX, B. J. WAINWRIGHT, K. BAKER, and J.
S. MATTICK. 1991. ‘Touchdown’ PCR to circumvent spurious priming during gene amplification. Nucleic Acids Res.
19:4008.
FU, Y. X. 1994. Estimating effective population size or mutation rate using the frequencies of mutations of various classes in a sample of DNA sequences. Genetics 138:1375–
1386.
———. 1996. Estimating the age of the common ancestor of
a DNA sample using the number of segregating sites. Genetics 144:829–838.
FU, Y. X., and W. H. LI. 1997. Estimating the age of the common ancestor of a sample of DNA sequences. Mol. Biol.
Evol. 14:195–199.
———. 1993. Statistical tests of neutrality of mutations. Genetics 133:693–709.
FULLERTON, S. M., J. BOND, J. A. SCHNEIDER, B. HAMILTON,
R. M. HARDING, A. J. BOYCE, and J. B. CLEGG. 2000. Polymorphism and divergence in the b-globin replication origin
initiation region. Mol. Biol. Evol. 17:179–188.
GRIFFITHS, R. C., and S. TAVARÉ. 1994. Ancestral inference in
population genetics. Stat. Sci. 9:307–319.
HARDING, R. M., S. M. FULLERTON, R. C. GRIFFITHS, J. BOND,
M. J. COX, J. A. SCHNEIDER, D. S. MOULIN, and J. B.
CLEGG. 1997. Archaic African and Asian lineages in the
genetic ancestry of modern humans. Am. J. Hum. Genet.
60:772–789.
HARRIS, E. E., and J. HEY. 1999. X chromosome evidence for
ancient human histories. Proc. Natl. Acad. Sci. USA 96:
3320–3324.
JARUZELSKA, J., E. ZIETKIEWICZ, M. BATZER, D. E. C. COLE,
J.-P. MOISAN, R. SCOZZARI, S. TAVARE, and D. LABUDA.
1999. Spatial and temporal distribution of the neutral polymorphisms in the last ZFX intron: analysis of the haplotype
structure and genealogy. Genetics 152:1091–1101.
JARUZELSKA, J., E. ZIETKIEWICZ, and D. LABUDA. 1999. Is selection responsible for the low level of variation in the last
intron of the ZFY locus? Mol. Biol. Evol. 16(11):1633–
1640. J. Mol. Evol. 21:58–71.
KAESSMANN, H., F. HEIßIG, A. VON HAESELER, and S. PÄÄBO.
1999. DNA sequence variation in a non-coding region of
low recombination on the human X chromosome. Nat. Genet. 22:78–81.
LI, W. H. 1997. Molecular evolution. Sinauer, Sunderland,
Mass.
LI, W. H., and L. SADLER. 1991. Low nucleotide diversity in
man. Genetics 129:513–523.
LI, W. H., C.-I. WU, and C.-C. LUO. 1984. Nonrandomness of
point mutation as reflected in nucleotide substitutions in
pseudogenes and its evolutionary implication. J. Mol. Evol.
21:58–71.
NICKERSON, D. A., S. L. TAYLOR, K. M. WEISS, A. G. CLARK,
R. G. HUTCHINSON, J. STENGARD, V. SALOMAA, E. VARTIAINEN, E. BOERWINKLE, and C. F. SING. 1998. DNA sequence diversity in a 9.7-kb region of the human lipoprotein
gene. Nat. Genet. 19:233–240.
RIEDER, M. J., S. L. TAYLOR, A. G. CLARK, and D. A. NICKERSON. 1999. Sequence variation in the human angiotensin
converting enzyme. Nat. Genet. 22:59–62.
ROZAS, J., and R. ROZAS. 1999. DnaSP version 3: an integrated
program for molecular population genetics and molecular
evolution analysis. Bioinformatics 15:174–175.
222
Yu et al.
STRINGER, C. B., and P. ANDREWS. 1988. Genetic and fossil
evidence for the origin of modern humans. Science 139:
1263–1268.
TAJIMA, F. 1983. Evolution relationship of DNA sequences in
finite populations. Genetics 105:437–460.
———. 1989. Statistical method for testing the neutral mutation
hypothesis by DNA polymorphism. Genetics 123:585–595.
TAKAHATA, N. 1993. Allelic genealogy and human evolution.
Mol. Biol. Evol. 10:2–22.
WATTERSON, G. A. 1975. On the number of segregation sites.
Theor. Popul. Biol. 7:256–276.
WHITFIELD, L. S., J. E. SULSTON, and P. N. GOODFELLOW.
1995. Sequence variation of the human Y chromosome. Nature 378:379–380.
ZHAO, Z., J. LI, Y.-X. FU et al. (13 co-authors). 2000. Worldwide DNA sequence variation in a 10 kb noncoding region
on human chromosome 22. Proc. Natl. Acad. Sci. USA 97:
11354–11358.
ZIETKIEWICZ, E., V. YOTOVA, M. JARNIK, et al. (11 co-authors).
1998. Genetic structure of the ancestral population of modern humans. J. Mol. Evol. 47:146–155.
KEITH CRANDALL, reviewing editor
Accepted October 9, 2000